School of Agriculture, Food and Ecosystem Sciences - Theses
Permanent URI for this collection
Search Results
1 - 2 of 2
-
ItemPredicting the grain protein concentration of wheat from non-destructive measurements of the crop at anthesis( 2005)Grain protein concentration is an important specification for wheat, which determines the quality grade and price received by growers. It is difficult to achieve target grain protein concentration in semi-arid southern Australia, because of the low and variable rainfall. Growers may benefit from being able to predict grain protein concentration before harvest, especially where there is a threshold or `window' requirement for a particular grade. Grain outside specifications could be forward sold into other grades while prices were good. Spatial predictions of grain protein concentration would allow the pattern of harvest to be managed to optimise profit. This thesis proposed a method for predicting grain protein concentration from non-destructive measurements of the crop (spikes, spikelets) at or after anthesis. The theoretical propositions underlying the method were then evaluated using data from nitrogen fertiliser experiments, data from the literature, and a simulation exercise. The proposed method was to estimate grain number from spike or spikelet number. Variance in grain number, together with the diminishing returns response of grain number to nitrogen, would then be used. to estimate maximum grain number. Maximum grain number would be linked to a unique `critical' grain protein concentration, from which grain protein concentration at other grain numbers could be estimated. Spike and spikelet number were counted throughout grain-filling in nitrogen fertiliser experiments to determine the importance of time of counting. The time of counting was important for absolute, but not relative spike and spikelet numbers: Spike and spikelet number varied, throughout grain-filling, but interactions with nitrogen treatments were rare. Inclusion of spikelets in counting was based on glume length, which interacted with time of counting. Spike death was frequently observed and occurred in proportion to post-anthesis growth, at 0.187(±0.018) spikes/g. The rate with respect to grain yield was similar, at 0.190(±0.038) spikes/g. An analysis of mass/number relationships between grain, spike and spikelet number, and crop and spike biomass at anthesis, showed that grain number was better related to spike biomass, and that spike and particularly spikelet number, were better related to crop biomass. Spikelet number changed at .a rate of between 6.6 and 9.3 spikelets/g biomass across 'a range of experiments; spike number changed at a rate between 0.14 and 0.62 spikes/g. The interrelationships showed grain number should be related to spikelets/spike, and proportion of crop biomass in the spike. The relationships, however, only existed in some experiments and were not universal. An alternative suggested by the analysis was use of spike number as a direct proxy for grain number (ie. assuming constant grains per spike). Spike number was tested as a proxy for grain number initially by analysing the components of variance of grain number across nitrogen, rotation and plant density experiments. Spike number was the main component of variance in grain number (59.8- 71.0% of log(variance)) in nitrogen experiments, with no significant covariance between spike number and grains per spike. Grains per spike and covariance were much greater components of variance in plant density experiments, and grains per spike and spike number were equal sources of variance in rotation experiments, with small positive covariance. Spike number would be an unbiased, but not perfect proxy for grain number when nitrogen was the main factor varying, but not for factors related to rotation or plant density. Spike number and crop biomass at anthesis were compared as estimators of grain number in nitrogen experiments, in an analysis of the nature of the responses to nitrogen fertiliser. Grain number as an estimator of grain yield was included in the analysis to understand the likely effect of using grain number rather than yield as a predictor of grain protein concentration. Crop biomass at anthesis, spike number and grain number all reached maxima at similar nitrogen fertiliser rates, but crop biomass at anthesis was a more precise estimator for the maximum rate required for grain number (RMSE of nitrogen for maximum, 2.4 kg N/ha vs. 26.4 kg N/ha). Grain number had a maximum consistently higher (+32.6±8.0 kg N/ha) than the maximum for yield. Once nitrogen fertiliser rates were corrected for the different maxima, grain number and yield had identical relative response rates to nitrogen. The response rates of crop biomass at anthesis and spike number were both related to the response rate of grain number by a power relationship with exponent 0.6. The lack of methods for anticipating phase differences caused by late nitrogen application and pre-anthesis water deficit will prevent exploitation of these relationships in all environments. The estimation of maximum spike number from its variance was simulated across the width of an air-seeder, using consistent variations in nitrogen fertiliser rate between tynes to drive variance in spike number. Nitrogen fertiliser was normally distributed. It was possible to extrapolate the variance/spike number relationship to estimate the maximum only where the slope of the relationship was negative. Slopes close to zero caused errors. of fitting, where the `signal' from the relationship was indistinguishable from the `noise' in estimating variance. This coincided with low (below 0.8) relative spike numbers and led to over-estimation of low relative spike numbers. Low spike number because of sub- or supra-optimal nitrogen could be distinguished by the second derivative of the fitted function, which was positive for supra-optimal nitrogen. There was no unique `critical' grain protein concentration (for maximum yield or grain number) in southeastern Australia, but there was a consistent relationship between `critical' grain protein concentration and grain weight. The relationship in terms of grain nitrogen content was a linear function of grain weight. The parameters also varied with genotype, and signed relative grain number, calculated as GRS=1-G/GMax for supraoptimal nitrogen, and GRS=G/GMax-1 for sub-optimal nitrogen, where G is grain number. The best estimation of grain nitrogen across genotypes was: Grain N (mg N/grain) = 0.317 + 1.00 x GRS + (0.0115 -0.0181 x GRS) x W, where W is grain weight in mg/grain. The root mean squared error of grain protein concentration estimated from this function was 0.91%. Grain weight would need to be estimated to estimate grain protein concentration. Errors due to grain weight had more effect at higher GRS, and at lower grain weight. The conclusion was that grain protein concentration may be predicted using crop biomass or spike number as a proxy for grain number. Predictions would be best in the absence of pre-anthesis water deficit or nitrogen applied after Zadoks 32. The predictions would be best for relative grain number greater than 0.8 at sub-optimal nitrogen, and for any relative grain number at supra-optimal nitrogen. A confidence interval could still be provided for grain protein concentration at lower relative grain numbers with sub-optimal nitrogen. Predictions would be most accurate if grain weight was reliably above 35 mg/grain.
-
ItemRoot and top growth of the wheat plant as affected by water supply( 1987)Two experiments were conducted to study the growth, particularly the root growth, of wheat in relation to moisture supply. One was conducted in an igloo house at the Mt Derrimut Field Station of the School of Agriculture and Forestry, University of Melbourne and the other in a glass house on the main campus. The plants were grown in a mixture of soil, sand and ligna peat contained in plastic bags to give columns about 0.8 m deep. In the first experiment the effects on both above-ground and below-ground growth of withholding water at different stages of growth and for different times were studied on three cultivars of wheat. In the second experiment effects of supplying different amounts of water were studied. Plants were harvested periodically for measurement of dry weights of the various plant parts, measurement of leaf area and root length and counts of tillers, heads, grains and roots. The soil columns were dismantled 20 cm at a time to enable the distribution of roots in the profile to be observed. Withholding water at any stage of growth and for any period reduced the above-ground dry matter at maturity; the longer the period of non-watering the greater the reduction. Reduction in above-ground dry matter resulted from reductions in number of tillers, leaf area and grain yield. Tillering was most affected by non-watering during early and mid-season growth, and yield by non-watering after anthesis, especially when it occurred during the first two weeks after anthesis. Non-watering after anthesis reduced yield mainly by reducing grain size (1000 grain weight). The quantity of recoverable roots, measured either by weight or by length, reached a maximum about the time of anthesis and then declined as roots died and were lost during harvest. The quantity of roots formed was reduced when non-watering was imposed during early or mid-growth. The distribution of roots in the profile was also affected by the watering treatments. Withholding water during early and mid-growth resulted in a greater proportion being located in the deeper soil zones. However, with the soil initially below field capacity, withholding water from sowing onwards resulted in shallow rooting presumably because an absolute lack of water limited the plants' ability to produce deep roots. Watering the soil to field capacity every three days in Experiment 2 resulted in less above-ground dry matter than watering to field capacity once a week. It resulted also in a greater proportion of the roots being in the upper part of the profile suggesting that root penetration of the lower part was inhibited by poor aeration as a consequence of overwatering. Watering to field capacity once a week resulted in less above-ground dry matter than giving half the amount of water needed to restore the whole soil column to field capacity. Moreover, the proportion of roots in the bottom zone of the rofile was less under the former treatment than under the latter suggesting that watering to field capacity once a week was causing some restriction of root growth in the bottom zone, presumably through poor aeration. These two facts taken together suggest that in this experiment watering to field capacity once a week even amounted to over-watering.